Markus Schumacher (Albert-Ludwigs-Universität Freiburg)

Markus Schumacher (Albert-Ludwigs-Universität Freiburg)IOP Workshop on VBF, 23 February 2011, Oxford

Higgs Boson Production via

Vector Boson Fusion in ATLAS

Experimental Challenges and Prospects

Signal rates for Higgs Boson Production Gluon gluon fusion (GGF)

Vektor boson fusion (VBF)

VBF searches: H2 photons, H 2 , t HWW 2(l+n )

VBF important contribution to the discovery potential

VBF key ingredient for investigation of Higgs profile (couplings, CP, ...)

Based on

LHC

H X

SE

C G

roup

M. Schumacher VBF Higgs boson searches with ATLAS: Challenges and Prospects Oxford, 23.2.2011

Signature of Vector Boson Fusion

W(Z)

W(Z)

signature:- 2 forward tag jets in opposite hemishperes with rapidity gap - Higgs decay products in

central detector between tag jets- no additional jets due to no colour flow in t channel- WW, : tt missing transverse energy due to neutrinos


Tagging of Forward Jets in VBF (H tt 14 TeV study)

jet reconstruction: (at CSC times cone R=0.4 now anti-kt R=0.4)

- high efficiency for tagging jets up to pseudorapidty of 4.8 (1 degree)

- fake rate only few %

- currently moderate sensitivity to pileup observed

(depending on noise suppression tool, cluster and jet algo used)


Detector Performance from Collision Data 2010

jet energy scale

jet reconstruction efficiency

jet energy resolutuoin

Anti-kt R=0.6

MET resolution


W e n candidate

M. Schumacher Higgs boson searches with ATLAS: Prospects and First Results Oxford, 22.2.2011

Z + Jets Production (shown Zee)

good description by MC generators

with matching of matrix element and parton shower at leading order

higher precision needed todiscrimniate quality of description

similar results for W+jets

after background subtraction and unfolding (jet pt>20GeV)


H : gg 14 TeV MC Study (CSC Book) inclusive analysis:

2 photons, pt >40 and 25 GeV mass window: MH+-1,4 sM

signal to background ratio: 1 to 38

significance with 10fb-1: 2.4

background from sidebands

fit exponential with nuisance

parameters: slope and norm.

MC 14 TeV

backgrounds:

irreducible: gj, jj, Zee

reducible: gg

using more information pt(Higgs)

etc. increase significance by 50%

signal: crystal ball + gaussian


H : gg 14 TeV MC Study (CSC Book)

VBF analysis:

- 2 photons, pt >50 (25) GeV - 2 jets,pt>40 (20) GeV with h1*h2<0, Dh>3.6 - photons btw. tagging jets - m jj >500 GeV - veto on additional jet with pt>20GeV and |h|<3.2 - mass window: MH+- 2GeV

accepted signal: GGF: 0.18 fb VBF: 0.79fb

accepted background: 1.95 fb (gg: 0.86fb, gj:0.42fb, jj:0.06fb, ggjj(EW): 0.59fb)

good signal to background ratio: 1 to 2

observation significance with 10fb-1: 2.0

background from sidebands

first data: only inclusive analysis

considered so far

MC 14 TeV

MC 7 TeV


H WWlnln with 35 pb-1 at 7 TeV

preselection:

- 2 pleptons (e,m), pt >20(15) GeV - Mll >15, |MZ-Mll|>10GeV (ee,mm)

- MET>30 GeV: em (ee,mm) - Dfll <1.3 (1.8) MH<170 (>=170GeV)

signal production via gluon fusion (MC@NLO) and VBF (SHERPA) considered backgrounds: diboson qqbar,ggWW,ZZ,WZ, W+jets, Z+jets, ttbar, single top

final cut on transverse mass 0.75 MH< <MH


HWWlnl : n Branching in Jet Multiplicites

uncertainty on fraction of signal events

0 jet: 10% 1 jet: 6% 2 jet:35%

evaluated by variation of renorm., factor. scales, PDFs, as in HNNLO

additional cuts for each jet multiplicity branch:

0jet: <30 GeV

1jet: veto on b-jet , <30GeV, Ztt veto,

2 jet: veto on b-jet, <30 GeV, Ztt veto, <80GeV

plus VBF cuts: mjj>500GeV, Dhjj>3.8, no 3rd central jet pt>25 GeV |h|<3.2

branch analysis in 0, 1 and 2 jets

- jets reconstructed with anti-kt

R=0.4 algorithm, pt>25 GeV, |h|<4.5

- associated to primary vertex PV

by requiring fraction of momentum from

tracks from PV to total jet track

momentum> 0.75 (for jets |h|<2.1)


dijet events in data:

no 2 add. jets with pt>20 GeV

Distributions and cut flow in 2 jet analyis


HWWlnl n : Selection Results

H+0 jets: H+1jet: H+2 jets:


Background Estimate from Data create control region CR for each individual background contribution extrapolate BG from CR to signal region SR with a factor obtained from MC pollution in control region constrained from other control regions via b factor from MC

consider exp. uncertainties ( E scale and resolution, tagging efficiencies, ...)

and theo. uncertainties (variation of renormalisation scales, ...)

as systematic uncertainty on a and b factors


First ATLAS Result in Higgs Boson Searches

Cross section limits:

54 (11,71) pb at MH=120 (160,200) GeV

Contribution from VBF marginal

1.2 x SM cross section

excluded at MH = 160 GeV

Limit calculation based on profile likelihood with 16% power constraint

Power contraint: if observed < expected -1 s then quote expected – 1 s


HWW MC: Sensitivity at 10 and 7 TeV

Limit calculation based on CLS method as used at LEP and TEVATRON

Importance of contribution from VBF

depends on luminosity.

Understanding of scaling with lumi

needs further studies.


MH=170 GeV

HWWe : nmn jet veto using track jets (14 TeV MC Study)

minimise influence of pile-up by using track jets associated to primary vertex

survival probability for jet veto (DR=0.5, cone): calo jets pt>20GeV |h|<4.8,

track jets >12.3 GeV |h|<2.5


Weak vector boson fusion H tt (14 teV MC Study)

background: reducible ---------------------------------------------- irreducible

kinematics, colour flow, … mass reconstruction


veto on additional jet with Pt>20 GeV and | h| <3.2 (ATLAS)

influence of pile up significant use of tracking information under investigation

Central Jet Veto

radiation in signal close to tagging jets rapidity gap QCD background (Z+jets,tt): additional central jets likely

EW QCD

D. Zeppenfeld et al., Phys.Rev.D54 (1996)6680 different color flow in EW and QCD processes


Mass Reconstruction: 14 TeV MC Study

mass resolution s M/M ~ 9 % dominated by missing transverse energy - 40% contrinbution of it from approximation - w/ pile up: 30% worse in 14 TeV MC study

in collinear approximation

despite four neutrinos

tt+W+jets

ZjjHiggs

Higgs boson on tail of Z peak mass resolution no easy sideband method more sophisticated method for background estimation needed


use jjZmm to model jjZttX as they have same topology

Zmm: signal free and high purity control sample selectable

apart from m and tX energy deposits same detector response

(including pileup, underlying event, noise, …)

Methodology:

1) select Zmm event in collision data

2) use 4-momenta of m as input for t decays

3) simulate ZttXY decay

4) replace cones around m in data event

by cones in simulated Ztt decay

on calorimeter cell level

„embedding“

5) re-reconstruct merged hybrid event

6) apply standard selection

Z tt background from data via „embedding“


mass distributions after embedding

Background Estimation from Data: Z tt (MC study)

- works for all tau lepton decay modes

- shape: prediction with better than 10% accuracy

- normalisation: i) from sideband

ii) nuisance parameter in statistics machinery

ii) theory prediction or measured Zll cross section


Daten Driven Estimate for H tt l had (Data) selection: 1 electron or muon pt>15 GeV 1 tau candidate pt>20 GeV

MET>20 GeV <30GeV

comparison of MC expectation and observed event yield

QCD jet expectation normalised to

MET<15 GeV control region

agreement between MC prediction and data

nevertheles aim for data driven BG estimate

(OS versus SS tau + lepton candidates)


Daten Driven Estimate for H tt l had. assumptions:

shape of visible mass distribution is the same for OS and SS events

ratio r between OS and SS events is the same in signal and control region

number of events in signal region (OS) can be expressed like:

define k as the deviation from r = 1 yields:

finally one gets:

obtained from data data MC MC


Determination of rQCD

missing transverse energy MET for OS and SS events after had. tau cut :

QCD control region missing transverse energy MET < 15 GeV (signal >20 GeV)

data: OS 139 SS 125 rqcd= 1.11+-0.14

use rQCD=1.00+-0.11+-0.14


Determination of rWW (kWW)

transverse mass for OS and SS events after MET cut :

W control region MT>50GeV (signal region<30GeV)

data: OS 15 (1.1 non WW BG from MC) 8 (0.5 non WW BG from MC)


Result of BG Estimation

nSSall: 16 +- 5

kwjet nssw+jet = 7.6+- 7.8

kother nother = 1.9 +- 0.5

sum: = 25 +- 9

Data driven MC prediction

observation: 29


Uncertainty on Signal Efficiency (14 TeV MC study)

dominant experimental systematic uncertainty: jet energy scale

seems ATLAS can do better than that

parton level uncertainties and parton-shower and underlying event model

10% indicate our belive that we will achieve this precision after tuning

and using better simulation tools for signal

(in 2008 comparsion of HERWIG, PYTHIA, SHERPA indicate 40% uncertainty)

determination of CJV survival probabilty from data appreciated


VBF Htt Sensitivity at 14 and 7 TeV

rescaled from 14 TeV MC study

for 1fb-1 at 7 TeV

assuming 20% uncertainty on

signal and Ztt background

and 50% on W+jets, tt, ...


Combined Projected Sensitivity at 7 TeV

expected 95% CL exclusion with 1fb-1 at 7 TeV from ~130 to ~ 450 GeV

low mass most difficult: dominated by Hgg

important contribution from H tt and Hbb


Conclusions

ATLAS detector performs very well, sometimes better than expected

VBF still a very promising channel for discovery and exclusion

exp. uncertainties are mostly under control:

- jet energy scale better than expected

- use of tracking information stababilises CJV aganist effects from pile-up

- missing energy reconstruction with pile up still improvable

need better understanding and modelling in MC event generators

(use SHERPA (w/ MENLOOPS) and POWHEG)

- jet multiplicities for braching of analysis and CJV

- determination of CJV survival probabilites for EW processes

also to be studied in the LHC Higgs cross section WG

(Sinead is the contact person for VBF production in ATLAS)


Estimating the CJV Efficiency from Dataknowledge of central jet veto efficiency needed for investigation of properties and optimisation of selection strategy

find and select samples with similar topology as VBF signal

with reasonable rate and signal-to-background ratio determine radiation pattern and transfer to Higgs signal process

directly or via tuning of MC generators

QCD

EW

D. Zeppenfeld et al., Phys.Rev.D54:6680-6689,1996

loose cuts:

EW/QCD = 1:7.2 sEW = 87 fb

tight cuts:

EW/QCD= 1.6 (2.9) sEW = 11(5) fb

the obvious candidate: jjZee+mm

after basic VBf cutscompeting QCD and EW contribution


CJV Efficiency from Data via a single top? (LH07)

signal: BG (mainly tt)~ 2 :1? similar radiation pattern ? similar topology selectable? what is influence of differences?

q q

b tW

W

b

n

e,m

Single top q q

q q

W

WH

t

t

VBF Htt


Uncertainty on signal efficiency (14 TeV MC study)


Combined Projected Sensitivity at 7 TeV (CLS method)

expected luminosity required for

95% CL exclusion

expected luminosity required for

95% CL exclusion

3 and 5 s observation

channels considered:

inclusive Hgg

VBF Htt ll, l had.

W(Z)H, Hbb (boosted)

HWWlnln (=0,1,2, jets)

HZZ4l

HZZllnn and llbb


One example: ttbar and WW BG in H+1 jet analysis

ttbar control region: replace b-veto by b-tag and drop cuts on mll, MT, Dfll

similar methids for all background in all jet mutiplicity bins

WW control region: mll >100 GeV and drop cuts on MT, Dfll

att + systematic uncertainty

aww + systematic uncertainty

combining both tables btt (pollution in WW control region)


Summery of sytematic uncertainties

Background extraction


Experimental uncertainties

Result of BG estimation

nSSall: 16 +- 5

kwjet nssw+jet = 7.6+- 7.8

kother nother = 1.9 +- 0.5

sum: = 25 +- 9

observation: 29


overview of systematic uncertainties

Data driven MC prediction

Determination of WW background

transverse mass and

visible mass shapes

agree within 10%

for OS and SS events

correction factor

for MT requiremnt


Markus Schumacher (Albert-Ludwigs-Universität Freiburg)

Documents

Transcript of Markus Schumacher (Albert-Ludwigs-Universität Freiburg)